Public reference workflows showing how I approach data automation, document extraction, scraping, reporting, validation, and AI-assisted review.
These repositories use synthetic, low-stakes examples, but the workflow patterns are real: messy inputs, structured outputs, validation checks, human-review cues, and maintainable handoff.
| Area | What it demonstrates | Typical client need |
|---|---|---|
| PDF/OCR and document extraction | Convert messy PDFs or scanned-like documents into structured, reviewable outputs | Extract tables, catalogues, forms, packets, or operational documents |
| Web scraping and data extraction | Collect heterogeneous website data, normalize records, and export structured datasets | Build repeatable data collection workflows from public or semi-structured sources |
| Spreadsheet and workbook automation | Sync, validate, and audit Excel/CSV workflows | Reduce manual spreadsheet work and create controlled reporting processes |
| AI-assisted workflows | Use LLMs for bounded extraction or classification with validation and human review | Add AI support without treating model output as automatically correct |
| Reporting automation | Convert structured data into repeatable PDF, HTML, Excel, or review-ready deliverables | Replace manual reporting with reproducible outputs |
| Research and geospatial pipelines | Process, document, and publish analytical or scientific datasets | Build reproducible research-grade data workflows |
-
PDF/OCR Catalog Review Workflow
Converts varied catalog-style PDFs into structured draft records, CSV/JSON/SQLite outputs, validation warnings, source evidence, and human-review packets. -
Supplier PDF to Excel/CSV Review Packet
Recovers structured rows from messy supplier-style PDFs, applies confidence scoring and validation flags, and produces an Excel/CSV handoff for review. -
Web Listings Extraction & Normalization Pipeline
Collects static, paginated, JavaScript-rendered, and cookie-gated listings into one normalized CSV/JSON catalog with an inspectable HTML QA report. -
PDF/DOCX Packet Structuring Workflow
Turns mixed document packets into source-cited run sheets with validation warnings, unresolved questions, and reviewer-facing outputs. -
AI-Assisted Email Intake Review Workflow
Extracts fields from informal emails with LangGraph/OpenAI, matches them against workbook rules, and writes review packets without auto-sending. -
Excel Formula Sync & Audit Workflow
Syncs reviewed formulas into an Excel-compatible workbook while rejecting unsafe or invalid rows and preserving an audit trail. -
AI-Assisted Publishing Review Workflow
Turns incomplete event/workshop requests into draft public copy, organizer digests, explicit review flags, and hold/publish decisions. -
CSV-to-PDF Automated Reporting Pipeline
Rebuilds a repeatable PDF packet from structured CSV data with validation, deterministic ordering, and stable layout.
The goal is not to showcase large production systems with private client data. The goal is to make my delivery style visible:
- clear problem framing;
- documented inputs and outputs;
- structured exports;
- validation and QA checks;
- reviewable intermediate files;
- reproducible workflows;
- maintainable handoff.
Python, R, ETL, data extraction, PDF/OCR workflows, web scraping, Excel/CSV automation, automated reporting, AI-assisted workflows, LangChain/LangGraph-style orchestration, validation, QA, reproducibility, and maintainable data pipelines.
I help clients turn messy data, documents, websites, spreadsheets, and research workflows into reliable analyses, automated pipelines, and decision-ready outputs.
My work combines practical implementation with statistical rigor, reproducible delivery, documentation, validation, and maintainable handover.
